20. • Making progress visible
• Delivering working software
• Collaboration towards shared goal
• Acting as Scrum Master
• Various roles in Feature Teams
Individuals and interactions
Co-located teams in Copenhagen
24. Responding to change
Taking on different roles and self organising
Breaking down
requirements
Developing
Product
Roadmaps
Facilitating
Prioritisation
Customer
Experience
Design
Scrum
Master
Feature
Team Coach
Observing
Customer Needs
Real
Options
Adapt to
Customer
Feedback
Managing
dependencies Managing
change
Business
Analyst
Product
Owner
25. Lessons from the revolutionary approach
Manage
stakeholder
expectations
Don’t attempt
to scale until
you’re ready
Defer
architectural
decisions
Legacy
dependencies
are painful
Only tacit
knowledge of
‘agile’...
26. 0
100
200
300
400
500
600
30
60
90
120
150
180
210
240
270
300
330
360
390
420
450
480
More
#Requirements
Days
Cycle time analysis
From Lightbulb (idea) to Live (in production)
Median = 150 days
27. From Revolution to Evolution...
New
Project, Platform, Team
Revolutionary
Existing
Setup, Platform, Team
Evolutionary
Lean
Product
Development
29. Mature practices they weren’t leveraging
Lean Product Development
Agile
SCRUM
Enterprise
Practices
Team
Practices
Project
Practices
XP Engineering
Practices
30. Selected Lean Enterprise practices for Maersk Line
First steps in improving the whole...
Contains 8 practices selected for Maersk Line:
1. Get to initial prioritisation faster
2. Improve prioritisation
3. Pull Requirements from Dynamic Priority List
4. Reduce size of requirements
5. Get to the point of writing code quickly
6. Actively manage Work-In-Progress (WIP)
7. Enable Faster Feedback
8. Enable smooth, sustainable flow
31. <1 week
Dynamic
Priority
List
Prioritise new
ideas quickly
Triage
Refine
Realise
Release
Prioritise by
knowing the
Cost of Delay
Manage
capacity with
a pull system
Increase
quality with
fast feedback
Get the point
of writing
code quickly
Break down
the work
Limit the Work in
Progress
Reduce batching
to smooth flow
Scope: Lightbulb... to Live
The end-to-end innovation process
32. 0
100
200
300
400
500
600
30
60
90
120
150
180
210
240
270
300
330
360
390
420
450
480
More
#Requirements
Days
Cycle time analysis
From Lightbulb (idea) to Live (in production)
Median = 150 days
During 2010
Med = 373 days
GCSS
33. State # RQ
Idea! 729
New 897
Being Drafted 416
Ready for Review 422
Ready for Guesstimation 181
Ready for Prioritization Analysis 2980 335
Ready for Estimation 68
Ready for Authorization 219
Authorized Estimation 502 215
Development Initiated 242
Development Complete Development 326 84
Testing (total of all states) Testing 395 395
On Hold Waiting 471 471
Fuzzy Front End
* Requirements work-in-progress (October 2010)
34. Go Live!
Dev & Test
(82 hrs)
Captured
PoC
(24 hrs)
18 weeks waiting
9 weeks waiting
11 weeks waiting
Wait waste = 38 weeks
Fuzzy Front End
35. Focus the upstream process
1. Get to initial prioritisation faster 5. Get to point of writing code quickly
<1 week
<2 weeks
Triage
Dynamic
Priority
List
Max 5
Refine
Pull to
coding
…
Dev
Buffer
Expect >10% attrition
otherwise upstream
process is too heavy
Don’t waste time
doing too much
analysis too early!
Don’t wastetime on low-value ideas
Quickly identify the
ideas that will be the
most profitable
36. Problem: Demand is unlimited
Consumerisation of I.T.
high expectations…
Most change is now
enabled by I.T. so they
need more
40. Benefits of using Cost of Delay
• ‘Less yelling and screaming’ data-driven, more visible
• Enables better trade-off decisions and increased ROI
• Handles dependencies between teams
• Changes the conversation…
Cutting I.T.
costs
Delivering
value quickly
Delivering
“on time”
M
H
42. Try this at home...
Ask each person on one of your project teams:
What would you estimate the
Cost of Delay
for this project to be?
43. System A: Value by quartile
$230,000/wk
$220/wk
Bottom
25%
Top 25% of RQs
$18,600/wk
Next 25%
$5,200/wk
Next 25%
Average $ Benefits
Per Requirement
44. $0
$200,000
$400,000
$600,000
$800,000
$1,000,000
$1,200,000
$1,400,000
$1,600,000
$1,800,000
$2,000,000
$2,200,000
$2,400,000
$2,600,000
$2,800,000
0
10
20
30
40
50
60
70
80
Requirements sorted by Cost of Delay
CostofDelay/week
A small number
of features have
a very high Cost
of Delay
System A: Value distribution
45. System B: Value distribution
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
0
10
20
30
40
50
60
80:20
Pareto Principle
“the vital few”
Requirements sorted by Cost of Delay
CostofDelay/week
46. <1 week
Dynamic
Priority
List
Prioritise new
ideas quickly
Triage
Refine
Realise
Release
Prioritise by
knowing the
Cost of Delay
Manage
capacity with a
pull system
Increase
quality with
fast feedback
Get the point
of writing code
quickly
Break down
the work
Limit the Work in
Progress
Reduce batching
to smooth flow
Reduce batching to smooth flow
47. Before: Feast and Famine
The effect of creating large release batches upstream
Requirements
S Des Dev T
S Des Dev T
S Des Dev T
S Des Dev T
R22
R23
R24
R25
Jan
201
Dev Dev Dev Dev
…Vendor team had ~10,000 hours of idle time in 2010
Development
Perspective:
Time
13 wks
50. <1 week
Dynamic
Priority
List
Prioritise new
ideas quickly
Triage
Refine
Realise
Release
Prioritise by
knowing the
Cost of Delay
Manage
capacity with
a pull system
Increase
quality with
fast feedback
Get the point
of writing code
quickly
Break down
the work
Limit the Work in
Progress
Reduce batching to
smooth flow
The end-to-end innovation process
Pull System
From Dynamic Priority List
51.
52. Fund the capacity to deliver change
...and make small adjustments over time
Time
Throughput
Mobilise
(3 months?)
Learning curve
(9 months?)
Stop!
Go!
53. Required flexible funding
3 potential models
Fund a given team size for a period of time
Fund small batches of requirements in advance
Fund individual requirements on demand
Time–based
Buffering
Just-in-Time
56. Predicting scope using data
Refine
Clarify WHW Check
DPL:
Priority
Q: Dev
Buffer
RQ-XXX
Realise
Dev Demo Tests
Production
18/07
14/07
07/07
05/07
27/06
23/06
04/07
14/08
15/08
12/07
05/08
03/08
18/07
14/07
02/08
11/09
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
RQ-XXX
Rn
Rn+1
Then, work backwards from fixed release dates to the option expiry date
Measure duration for each step.
57. Variable
Typical
measures
Usual
outcomes
Lean-‐Agile
alternaBves
Time
Delivering
on
a
predicted
date
Incen;vises
hidden
;me
buffers
and
slower
delivery
Maximise
speed
in
geGng
to
the
point
where
value
starts
to
be
realised
Scope
Delivering
all
of
the
originally
predicted
scope
Incen;vises
gold
pla;ng
and
discourages
exploita;on
of
learning.
Minimize
size
of
work
packages
to
maximize
both
learning
and
early
release
of
value
Cost
Delivering
at
or
below
a
predicted
development
cost
Incen;vises
hidden
cost
con;ngencies,
pushing
costs
up.
Maximize
value
delivered
(trade
development
cost
against
the
opportunity
cost
of
delay)
Quality
Delivering
changes
with
zero
down;me
and
no
errors
Resistance
to
making
any
changes.
Overinvestment
in
tes;ng
&
documenta;on.
Shorten
feedback
cycles
at
many
levels
(coding,
defects…)
Key Performance Measures for IT
58. Refine
<1 week
Dynamic
Priority
List
Prioritise new
ideas quickly
Triage
Realise
Release
Prioritise by
knowing the
Cost of Delay
Manage
capacity with a
pull system
Increase
quality with
fast feedback
Get the point
of writing code
quickly
Break down
the work
Limit the Work in
Progress
Reduce batching to
smooth flow
Break down the work
60. <1 week
Dynamic
Priority
List
Prioritise new
ideas quickly
Triage
Refine
Realise
Release
Prioritise by
knowing the
Cost of Delay
Manage
capacity with a
pull system
Increase
quality with
fast feedback
Get the point
of writing code
quickly
Break down
the work
Limit the Work
in Progress
Reduce batching to
smooth flow
Constraining Work-in-Progress
62. 190
#Requirements*
46
6.0
5.2
6.1
7.9
8.8
6.4
7.1
Rel 19-22
R23
R24
R25
R26
R27
R28
Work-in-Progress. Controlled.
Oct 2010
Jan 2012
76%
…whilst at least maintaining throughput
Reduced
wait waste
Improved
visibility
*”Authorized” to “Launched”
Guesstimate points/week
63. <1 week
Dynamic
Priority
List
Prioritise new
ideas quickly
Triage
Refine
Realise
Release
Prioritise by
knowing the
Cost of Delay
Manage
capacity with a
pull system
Increase
quality with
fast feedback
Get the point
of writing code
quickly
Break down
the work
Limit the Work in
Progress
Reduce batching to
smooth flow
‘Quality’ requires feedback
64. Faster Feedback
Eight Standard Measures
Requirement
captured
Requirement
validated
Started
coding
Integrated
& built
Completed
coding
Accepted
for launch
Launched
in production
Feasible Demonstrated
Accepted
Launched
Code
complete
Feature complete
Require-
ments
Release
candidate
Code
Launchable
65. Faster Feedback
Comparing GCSS with the X-leap on the Eight Measures
All times are in days
Action:
HOAT Forward
Action:
Faster SPT
66. <1 week
Dynamic
Priority
List
Prioritise new
ideas quickly
Triage
Refine
Realise
Release
Prioritise by
knowing the
Cost of Delay
Manage
capacity with a
pull system
Increase
quality with
fast feedback
Get the point
of writing code
quickly
Break down
the work
Limit the Work in
Progress
Reduce batching to
smooth flow
What about the outcomes?
67. Faster Delivery
How do we know the changes are an improvement?
Half
the time60
104
168
208
FACT
GCSS
90
150
Target
All
Apps
days
days
days
days
SAP
For GCSS : Prioritised to Live
For FACT : Idea to Live
68. Better Quality
How do we know the changes are an improvement?
8.2
11.2
2
1
2.2
0.3
Defects Delays Patches
88%
85%
80%
69. More Value
How do we know the changes are an improvement?
$26.30
$44.80
GCSS
FACT
MLIT Average
(Before)
Benefits per dollar invested
$4.10
(After)
70. And... delivering Cheaper
Not what we were aiming for, but reducing waste has led to...
$82.8
$75.6
Cost per hour
6
7.3
Throughput
22%
9%
The data is for GCSS
Throughput is measured as guesstimate point/week
71. “Absolutely
worth it”
People love it!
“Less yelling
& screaming”
“Smaller and clear changes
delivered faster”
“Fewer defects
in a release to handle”
“Daily calls provide
good visibility
of changes”
“We have not had such a smooth
launch since Release 16 –
I thought my phone had
stopped working”
“It
de
eff
wo
72. Learnings from an evolutionary approach
• It’s possible to make a huge impact on cycle time
without changing engineering practices
• Every team is like a separate company
• Long lead time to get to ‘kick off’
• It’s easier to sell when the process is knowingly broken
• Two steps forward one step backwards
• Organizational uncertainty is a challenge
• Huge push back for WIP limits from some
• It’s all about stakeholder management (70%)
• Systemic issues block adoption…
73. What went well?
• Enthusiastic portfolio managers make it easier to get buy-in
from the business partners
• People with right mindset and background in the team make a
difference
• People in the field get it; Practices work well on team level
• All parts of the organisation were on our Steering Group
• Combination of Consultants+Maersk resources as Lean-Agile
coaches
• COD adoption went viral after the pilot
• Presenting an engagement plan and defining the coach’s role
upfront
• Kanban boards and Standups go viral (but..)
74. What could have been better?
• Organisational changes shifted the entire stakeholder map
• We could have brought management closer to the work
• Introduce training on the principles early on to embed the
mindset
• Mandate to change the organisation and process
• Set up champions network early on
75. What still puzzles us?
• Success with legacy changed the perception that ‘agile works
with greenfield’
• How to predict the value for a pipeline over the longer term
• How to approach teams that are on a likely path to failure
• How to encourage the organisation as a whole to learn
• How to get senior managers to lead the transformation
• How to change the culture of an organisation
• How to explain Cumulative Flow Diagrams to people better
76. Bottom of the iceberg
It’s not just process…
Visible change
Hidden
Culture & Mindset
Change
on:
le’
Behaviours
Customs
Language
Values
Traditions
Beliefs
Stereotypes
Taboos
Org chart
Processes
Roles
Tools
77. Final piece of advice
• Don’t give up
• Avoid Agile vs. Waterfall > Outcomes
• Look at the whole end to end
• Make the problem visible
• Start with the principles
• Study your stakeholders
• Appeal to hearts & minds
79. If you’re not moving at the
speed of the marketplace
you're already dead
…you just
haven't stopped
breathing yet
Jack Welch
CEO of General Electric 1981-2001
“
”
80. Maersk Line Experience report and lots of other material
BLACK SWAN FARMING.COM
Want more?
ozlem@agileatheart.com
@OzzieYuce
Özlem Yüce